Datasets:

undertheseanlp
/

UVN-1

"""
Run all news scrapers to collect 1000 articles from each of 6 Vietnamese newspapers.
Total: 6000 articles
Features:
- 10 parallel workers
- Progress display
- Saves crawled URLs to avoid duplicates
"""

import json
import time
import random
import requests
from bs4 import BeautifulSoup
from pathlib import Path
from concurrent.futures import ThreadPoolExecutor, as_completed
from threading import Lock
import sys
import urllib3

# Suppress SSL warnings
urllib3.disable_warnings(urllib3.exceptions.InsecureRequestWarning)

DATA_DIR = Path(__file__).parent.parent / "data"
CRAWLED_FILE = DATA_DIR / "crawled_urls.json"
TARGET_PER_SOURCE = 1000
NUM_WORKERS = 10

# Lock for thread-safe operations
file_lock = Lock()
progress_lock = Lock()

# All newspaper configurations
NEWSPAPERS = {
    "vnexpress": {
        "base_url": "https://vnexpress.net",
        "categories": {
            "thoi-su": "Thời sự",
            "the-gioi": "Thế giới",
            "kinh-doanh": "Kinh doanh",
            "giai-tri": "Giải trí",
            "the-thao": "Thể thao",
            "phap-luat": "Pháp luật",
            "giao-duc": "Giáo dục",
            "suc-khoe": "Sức khỏe",
            "doi-song": "Đời sống",
            "khoa-hoc": "Khoa học",
        },
        "page_url": lambda base, cat, page: f"{base}/{cat}" if page == 1 else f"{base}/{cat}-p{page}",
        "list_selector": ["article.item-news h3.title-news a", "h3.title-news a"],
        "title_selector": ["h1.title-detail"],
        "desc_selector": ["p.description"],
        "content_selector": ["article.fck_detail"],
        "content_p_selector": "p.Normal",
        "date_selector": ["span.date"],
    },
    "dantri": {
        "base_url": "https://dantri.com.vn",
        "categories": {
            "xa-hoi": "Xã hội",
            "the-gioi": "Thế giới",
            "kinh-doanh": "Kinh doanh",
            "giai-tri": "Giải trí",
            "the-thao": "Thể thao",
            "phap-luat": "Pháp luật",
            "giao-duc": "Giáo dục",
            "suc-khoe": "Sức khỏe",
            "doi-song": "Đời sống",
            "khoa-hoc-cong-nghe": "Khoa học",
        },
        "page_url": lambda base, cat, page: f"{base}/{cat}/trang-{page}.htm",
        "list_selector": ["article.article-item h3.article-title a", "h3.article-title a"],
        "title_selector": ["h1.title-page", "h1.e-magazine__title"],
        "desc_selector": ["h2.singular-sapo", "div.singular-sapo"],
        "content_selector": ["div.singular-content", "div.e-magazine__body"],
        "content_p_selector": "p",
        "date_selector": ["time.author-time"],
    },
    "thanhnien": {
        "base_url": "https://thanhnien.vn",
        "categories": {
            "thoi-su": "Thời sự",
            "the-gioi": "Thế giới",
            "kinh-te": "Kinh doanh",
            "giai-tri": "Giải trí",
            "the-thao": "Thể thao",
            "phap-luat": "Pháp luật",
            "giao-duc": "Giáo dục",
            "suc-khoe": "Sức khỏe",
            "doi-song": "Đời sống",
            "cong-nghe": "Khoa học",
        },
        "page_url": lambda base, cat, page: f"{base}/{cat}.htm" if page == 1 else f"{base}/{cat}/trang-{page}.htm",
        "list_selector": ["a[title]", "h3 a", "h2 a"],
        "title_selector": ["h1.detail-title", "h1.details__headline", "h1"],
        "desc_selector": ["h2.detail-sapo", "div.detail-sapo", "p.sapo"],
        "content_selector": ["div.detail-content", "div.details__content", "div.content"],
        "content_p_selector": "p",
        "date_selector": ["div.detail-time", "time", "span.time"],
    },
    "tuoitre": {
        "base_url": "https://tuoitre.vn",
        "categories": {
            "thoi-su": "Thời sự",
            "the-gioi": "Thế giới",
            "kinh-doanh": "Kinh doanh",
            "giai-tri": "Giải trí",
            "the-thao": "Thể thao",
            "phap-luat": "Pháp luật",
            "giao-duc": "Giáo dục",
            "suc-khoe": "Sức khỏe",
            "xe": "Xe",
            "nhip-song-tre": "Đời sống",
        },
        "page_url": lambda base, cat, page: f"{base}/{cat}.htm" if page == 1 else f"{base}/{cat}/trang-{page}.htm",
        "list_selector": ["h3 a", "a.box-category-link-title", "a[title]"],
        "title_selector": ["h1.detail-title", "h1.article-title", "h1"],
        "desc_selector": ["h2.detail-sapo", "p.detail-sapo", "p.sapo"],
        "content_selector": ["div.detail-content", "div#main-detail-body", "div.content"],
        "content_p_selector": "p",
        "date_selector": ["div.detail-time", "span.date-time", "time"],
    },
    "tienphong": {
        "base_url": "https://tienphong.vn",
        "categories": {
            "xa-hoi": "Xã hội",
            "the-gioi": "Thế giới",
            "kinh-te": "Kinh doanh",
            "giai-tri": "Giải trí",
            "the-thao": "Thể thao",
            "phap-luat": "Pháp luật",
            "giao-duc": "Giáo dục",
            "suc-khoe": "Sức khỏe",
            "gioi-tre": "Đời sống",
            "cong-nghe": "Khoa học",
        },
        "page_url": lambda base, cat, page: f"{base}/{cat}.html" if page == 1 else f"{base}/{cat}/trang-{page}.html",
        "list_selector": ["a[href*='post']", "h2 a", "h3 a"],
        "title_selector": ["h1.article__title", "h1.cms-title", "h1"],
        "desc_selector": ["h2.article__sapo", "div.article__sapo", "p.sapo"],
        "content_selector": ["div.article__body", "div.cms-body", "div.content"],
        "content_p_selector": "p",
        "date_selector": ["time.article__time", "span.cms-date", "time"],
    },
    "nguoilaodong": {
        "base_url": "https://nld.com.vn",
        "categories": {
            "thoi-su": "Thời sự",
            "the-gioi": "Thế giới",
            "kinh-te": "Kinh doanh",
            "van-hoa-giai-tri": "Giải trí",
            "the-thao": "Thể thao",
            "phap-luat": "Pháp luật",
            "giao-duc": "Giáo dục",
            "suc-khoe": "Sức khỏe",
            "cong-doan": "Công đoàn",
            "cong-nghe": "Khoa học",
        },
        "page_url": lambda base, cat, page: f"{base}/{cat}.htm" if page == 1 else f"{base}/{cat}/trang{page}.htm",
        "list_selector": ["h3 a", "h2 a", "a[title]"],
        "title_selector": ["h1.title-detail", "h1.detail-title", "h1"],
        "desc_selector": ["p.sapo-detail", "div.sapo-detail", "p.sapo"],
        "content_selector": ["div.content-detail", "div.detail-content", "div.content"],
        "content_p_selector": "p",
        "date_selector": ["span.date-detail", "time.time-detail", "time"],
    },
}

HEADERS = {
    "User-Agent": "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36"
}


def load_crawled_urls() -> set:
    """Load previously crawled URLs."""
    if CRAWLED_FILE.exists():
        with open(CRAWLED_FILE, "r") as f:
            return set(json.load(f))
    return set()


def save_crawled_urls(urls: set):
    """Save crawled URLs to file."""
    with file_lock:
        DATA_DIR.mkdir(parents=True, exist_ok=True)
        with open(CRAWLED_FILE, "w") as f:
            json.dump(list(urls), f)


def select_element(soup, selectors):
    """Try multiple selectors and return first match."""
    for selector in selectors:
        elem = soup.select_one(selector)
        if elem:
            return elem
    return None


def is_article_url(url: str, base_url: str) -> bool:
    """Check if URL is likely an article (not category/image)."""
    if not url or not url.startswith(base_url):
        return False
    # Must have file extension or be a long enough path
    if ".htm" in url or ".html" in url or ".tpo" in url:
        # Exclude category pages (short paths)
        path = url.replace(base_url, "")
        # Article URLs typically have longer slugs with IDs
        return len(path) > 30 or any(c.isdigit() for c in path)
    return False


def get_article_urls(source: str, config: dict, crawled_urls: set) -> list[dict]:
    """Get article URLs from all categories."""
    urls = []
    base_url = config["base_url"]
    seen_urls = set()

    for cat_slug, cat_name in config["categories"].items():
        print(f"  [{source}] Category: {cat_name}", flush=True)
        cat_urls = 0

        for page in range(1, 30):
            url = config["page_url"](base_url, cat_slug, page)

            try:
                response = requests.get(url, timeout=30, headers=HEADERS, verify=False)
                if response.status_code != 200:
                    break
                soup = BeautifulSoup(response.text, "html.parser")

                page_urls = 0
                for selector in config["list_selector"]:
                    articles = soup.select(selector)
                    for article in articles:
                        href = article.get("href", "")
                        if href:
                            full_url = href if href.startswith("http") else f"{base_url}{href}"
                            if is_article_url(full_url, base_url) and full_url not in crawled_urls and full_url not in seen_urls:
                                seen_urls.add(full_url)
                                urls.append({
                                    "url": full_url,
                                    "category": cat_name,
                                    "source": source
                                })
                                page_urls += 1
                                cat_urls += 1

                if page_urls == 0:
                    break

                if len(urls) >= TARGET_PER_SOURCE * 2:
                    break

                time.sleep(random.uniform(0.2, 0.5))

            except Exception as e:
                print(f"  [{source}] Error page {page}: {e}", flush=True)
                break

        print(f"  [{source}] {cat_name}: {cat_urls} URLs", flush=True)

        if len(urls) >= TARGET_PER_SOURCE * 2:
            break

    return urls


def scrape_article(item: dict, config: dict) -> dict | None:
    """Scrape content from a single article."""
    url = item["url"]

    try:
        response = requests.get(url, timeout=30, headers=HEADERS, verify=False)
        response.raise_for_status()
        soup = BeautifulSoup(response.text, "html.parser")

        # Get title
        title_elem = select_element(soup, config["title_selector"])
        title = title_elem.get_text(strip=True) if title_elem else ""

        # Get description
        desc_elem = select_element(soup, config["desc_selector"])
        description = desc_elem.get_text(strip=True) if desc_elem else ""

        # Get content
        content_elem = select_element(soup, config["content_selector"])
        if content_elem:
            paragraphs = content_elem.select(config["content_p_selector"])
            content = "\n\n".join(p.get_text(strip=True) for p in paragraphs if p.get_text(strip=True))
        else:
            content = ""

        # Get publish date - REQUIRED
        date_elem = select_element(soup, config["date_selector"])
        publish_date = date_elem.get_text(strip=True) if date_elem else ""

        # Must have content and publish_date
        if not content or len(content) < 100 or not publish_date:
            return None

        return {
            "source": item["source"],
            "url": url,
            "category": item["category"],
            "title": title,
            "description": description,
            "content": content,
            "publish_date": publish_date
        }

    except Exception:
        return None


def scrape_source(source: str, config: dict, crawled_urls: set, all_crawled: set):
    """Scrape articles from a single source."""
    print(f"\n{'='*60}")
    print(f"[{source.upper()}] Starting...")
    print("="*60)

    # Get URLs
    print(f"[{source}] Fetching article URLs...", flush=True)
    article_urls = get_article_urls(source, config, crawled_urls)
    print(f"[{source}] Found {len(article_urls)} new URLs")

    if not article_urls:
        return []

    articles = []
    completed = 0

    def process_article(item):
        return scrape_article(item, config)

    with ThreadPoolExecutor(max_workers=NUM_WORKERS) as executor:
        futures = {executor.submit(process_article, item): item for item in article_urls}

        for future in as_completed(futures):
            item = futures[future]
            completed += 1

            try:
                article = future.result()
                if article:
                    articles.append(article)
                    with progress_lock:
                        all_crawled.add(item["url"])

                    # Progress display
                    if len(articles) % 10 == 0:
                        print(f"[{source}] Progress: {len(articles)}/{TARGET_PER_SOURCE} articles | Processed: {completed}/{len(article_urls)}", flush=True)

                    if len(articles) >= TARGET_PER_SOURCE:
                        break
            except Exception:
                pass

    print(f"[{source}] Completed: {len(articles)} articles")
    return articles[:TARGET_PER_SOURCE]


def main():
    print("="*60)
    print("UVN Dataset Scraper")
    print(f"Target: 6 newspapers x {TARGET_PER_SOURCE} articles = {6*TARGET_PER_SOURCE} total")
    print(f"Workers: {NUM_WORKERS}")
    print("="*60)

    # Load previously crawled URLs
    crawled_urls = load_crawled_urls()
    print(f"Previously crawled URLs: {len(crawled_urls)}")

    all_crawled = set(crawled_urls)
    all_articles = []

    for source, config in NEWSPAPERS.items():
        try:
            articles = scrape_source(source, config, crawled_urls, all_crawled)
            all_articles.extend(articles)

            # Save progress after each source
            save_crawled_urls(all_crawled)

            # Save articles for this source
            source_dir = DATA_DIR / source
            source_dir.mkdir(parents=True, exist_ok=True)
            output_file = source_dir / f"{source}_articles.json"
            with open(output_file, "w", encoding="utf-8") as f:
                json.dump(articles, f, ensure_ascii=False, indent=2)
            print(f"[{source}] Saved to: {output_file}")

        except KeyboardInterrupt:
            print("\nStopped by user. Saving progress...")
            save_crawled_urls(all_crawled)
            break
        except Exception as e:
            print(f"[{source}] Error: {e}")
            continue

    # Save all articles
    all_file = DATA_DIR / "all_articles.json"
    with open(all_file, "w", encoding="utf-8") as f:
        json.dump(all_articles, f, ensure_ascii=False, indent=2)

    print("\n" + "="*60)
    print("SUMMARY")
    print("="*60)
    print(f"Total articles: {len(all_articles)}")
    print(f"Total crawled URLs: {len(all_crawled)}")
    print(f"Saved to: {all_file}")


if __name__ == "__main__":
    main()

	"""
	Run all news scrapers to collect 1000 articles from each of 6 Vietnamese newspapers.
	Total: 6000 articles

	Features:
	- 10 parallel workers
	- Progress display
	- Saves crawled URLs to avoid duplicates
	"""

	import json
	import time
	import random
	import requests
	from bs4 import BeautifulSoup
	from pathlib import Path
	from concurrent.futures import ThreadPoolExecutor, as_completed
	from threading import Lock
	import sys
	import urllib3

	# Suppress SSL warnings
	urllib3.disable_warnings(urllib3.exceptions.InsecureRequestWarning)

	DATA_DIR = Path(__file__).parent.parent / "data"
	CRAWLED_FILE = DATA_DIR / "crawled_urls.json"
	TARGET_PER_SOURCE = 1000
	NUM_WORKERS = 10

	# Lock for thread-safe operations
	file_lock = Lock()
	progress_lock = Lock()

	# All newspaper configurations
	NEWSPAPERS = {
	"vnexpress": {
	"base_url": "https://vnexpress.net",
	"categories": {
	"thoi-su": "Thời sự",
	"the-gioi": "Thế giới",
	"kinh-doanh": "Kinh doanh",
	"giai-tri": "Giải trí",
	"the-thao": "Thể thao",
	"phap-luat": "Pháp luật",
	"giao-duc": "Giáo dục",
	"suc-khoe": "Sức khỏe",
	"doi-song": "Đời sống",
	"khoa-hoc": "Khoa học",
	},
	"page_url": lambda base, cat, page: f"{base}/{cat}" if page == 1 else f"{base}/{cat}-p{page}",
	"list_selector": ["article.item-news h3.title-news a", "h3.title-news a"],
	"title_selector": ["h1.title-detail"],
	"desc_selector": ["p.description"],
	"content_selector": ["article.fck_detail"],
	"content_p_selector": "p.Normal",
	"date_selector": ["span.date"],
	},
	"dantri": {
	"base_url": "https://dantri.com.vn",
	"categories": {
	"xa-hoi": "Xã hội",
	"the-gioi": "Thế giới",
	"kinh-doanh": "Kinh doanh",
	"giai-tri": "Giải trí",
	"the-thao": "Thể thao",
	"phap-luat": "Pháp luật",
	"giao-duc": "Giáo dục",
	"suc-khoe": "Sức khỏe",
	"doi-song": "Đời sống",
	"khoa-hoc-cong-nghe": "Khoa học",
	},
	"page_url": lambda base, cat, page: f"{base}/{cat}/trang-{page}.htm",
	"list_selector": ["article.article-item h3.article-title a", "h3.article-title a"],
	"title_selector": ["h1.title-page", "h1.e-magazine__title"],
	"desc_selector": ["h2.singular-sapo", "div.singular-sapo"],
	"content_selector": ["div.singular-content", "div.e-magazine__body"],
	"content_p_selector": "p",
	"date_selector": ["time.author-time"],
	},
	"thanhnien": {
	"base_url": "https://thanhnien.vn",
	"categories": {
	"thoi-su": "Thời sự",
	"the-gioi": "Thế giới",
	"kinh-te": "Kinh doanh",
	"giai-tri": "Giải trí",
	"the-thao": "Thể thao",
	"phap-luat": "Pháp luật",
	"giao-duc": "Giáo dục",
	"suc-khoe": "Sức khỏe",
	"doi-song": "Đời sống",
	"cong-nghe": "Khoa học",
	},
	"page_url": lambda base, cat, page: f"{base}/{cat}.htm" if page == 1 else f"{base}/{cat}/trang-{page}.htm",
	"list_selector": ["a[title]", "h3 a", "h2 a"],
	"title_selector": ["h1.detail-title", "h1.details__headline", "h1"],
	"desc_selector": ["h2.detail-sapo", "div.detail-sapo", "p.sapo"],
	"content_selector": ["div.detail-content", "div.details__content", "div.content"],
	"content_p_selector": "p",
	"date_selector": ["div.detail-time", "time", "span.time"],
	},
	"tuoitre": {
	"base_url": "https://tuoitre.vn",
	"categories": {
	"thoi-su": "Thời sự",
	"the-gioi": "Thế giới",
	"kinh-doanh": "Kinh doanh",
	"giai-tri": "Giải trí",
	"the-thao": "Thể thao",
	"phap-luat": "Pháp luật",
	"giao-duc": "Giáo dục",
	"suc-khoe": "Sức khỏe",
	"xe": "Xe",
	"nhip-song-tre": "Đời sống",
	},
	"page_url": lambda base, cat, page: f"{base}/{cat}.htm" if page == 1 else f"{base}/{cat}/trang-{page}.htm",
	"list_selector": ["h3 a", "a.box-category-link-title", "a[title]"],
	"title_selector": ["h1.detail-title", "h1.article-title", "h1"],
	"desc_selector": ["h2.detail-sapo", "p.detail-sapo", "p.sapo"],
	"content_selector": ["div.detail-content", "div#main-detail-body", "div.content"],
	"content_p_selector": "p",
	"date_selector": ["div.detail-time", "span.date-time", "time"],
	},
	"tienphong": {
	"base_url": "https://tienphong.vn",
	"categories": {
	"xa-hoi": "Xã hội",
	"the-gioi": "Thế giới",
	"kinh-te": "Kinh doanh",
	"giai-tri": "Giải trí",
	"the-thao": "Thể thao",
	"phap-luat": "Pháp luật",
	"giao-duc": "Giáo dục",
	"suc-khoe": "Sức khỏe",
	"gioi-tre": "Đời sống",
	"cong-nghe": "Khoa học",
	},
	"page_url": lambda base, cat, page: f"{base}/{cat}.html" if page == 1 else f"{base}/{cat}/trang-{page}.html",
	"list_selector": ["a[href*='post']", "h2 a", "h3 a"],
	"title_selector": ["h1.article__title", "h1.cms-title", "h1"],
	"desc_selector": ["h2.article__sapo", "div.article__sapo", "p.sapo"],
	"content_selector": ["div.article__body", "div.cms-body", "div.content"],
	"content_p_selector": "p",
	"date_selector": ["time.article__time", "span.cms-date", "time"],
	},
	"nguoilaodong": {
	"base_url": "https://nld.com.vn",
	"categories": {
	"thoi-su": "Thời sự",
	"the-gioi": "Thế giới",
	"kinh-te": "Kinh doanh",
	"van-hoa-giai-tri": "Giải trí",
	"the-thao": "Thể thao",
	"phap-luat": "Pháp luật",
	"giao-duc": "Giáo dục",
	"suc-khoe": "Sức khỏe",
	"cong-doan": "Công đoàn",
	"cong-nghe": "Khoa học",
	},
	"page_url": lambda base, cat, page: f"{base}/{cat}.htm" if page == 1 else f"{base}/{cat}/trang{page}.htm",
	"list_selector": ["h3 a", "h2 a", "a[title]"],
	"title_selector": ["h1.title-detail", "h1.detail-title", "h1"],
	"desc_selector": ["p.sapo-detail", "div.sapo-detail", "p.sapo"],
	"content_selector": ["div.content-detail", "div.detail-content", "div.content"],
	"content_p_selector": "p",
	"date_selector": ["span.date-detail", "time.time-detail", "time"],
	},
	}

	HEADERS = {
	"User-Agent": "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36"
	}


	def load_crawled_urls() -> set:
	"""Load previously crawled URLs."""
	if CRAWLED_FILE.exists():
	with open(CRAWLED_FILE, "r") as f:
	return set(json.load(f))
	return set()


	def save_crawled_urls(urls: set):
	"""Save crawled URLs to file."""
	with file_lock:
	DATA_DIR.mkdir(parents=True, exist_ok=True)
	with open(CRAWLED_FILE, "w") as f:
	json.dump(list(urls), f)


	def select_element(soup, selectors):
	"""Try multiple selectors and return first match."""
	for selector in selectors:
	elem = soup.select_one(selector)
	if elem:
	return elem
	return None


	def is_article_url(url: str, base_url: str) -> bool:
	"""Check if URL is likely an article (not category/image)."""
	if not url or not url.startswith(base_url):
	return False
	# Must have file extension or be a long enough path
	if ".htm" in url or ".html" in url or ".tpo" in url:
	# Exclude category pages (short paths)
	path = url.replace(base_url, "")
	# Article URLs typically have longer slugs with IDs
	return len(path) > 30 or any(c.isdigit() for c in path)
	return False


	def get_article_urls(source: str, config: dict, crawled_urls: set) -> list[dict]:
	"""Get article URLs from all categories."""
	urls = []
	base_url = config["base_url"]
	seen_urls = set()

	for cat_slug, cat_name in config["categories"].items():
	print(f" [{source}] Category: {cat_name}", flush=True)
	cat_urls = 0

	for page in range(1, 30):
	url = config["page_url"](base_url, cat_slug, page)

	try:
	response = requests.get(url, timeout=30, headers=HEADERS, verify=False)
	if response.status_code != 200:
	break
	soup = BeautifulSoup(response.text, "html.parser")

	page_urls = 0
	for selector in config["list_selector"]:
	articles = soup.select(selector)
	for article in articles:
	href = article.get("href", "")
	if href:
	full_url = href if href.startswith("http") else f"{base_url}{href}"
	if is_article_url(full_url, base_url) and full_url not in crawled_urls and full_url not in seen_urls:
	seen_urls.add(full_url)
	urls.append({
	"url": full_url,
	"category": cat_name,
	"source": source
	})
	page_urls += 1
	cat_urls += 1

	if page_urls == 0:
	break

	if len(urls) >= TARGET_PER_SOURCE * 2:
	break

	time.sleep(random.uniform(0.2, 0.5))

	except Exception as e:
	print(f" [{source}] Error page {page}: {e}", flush=True)
	break

	print(f" [{source}] {cat_name}: {cat_urls} URLs", flush=True)

	if len(urls) >= TARGET_PER_SOURCE * 2:
	break

	return urls


	def scrape_article(item: dict, config: dict) -> dict \| None:
	"""Scrape content from a single article."""
	url = item["url"]

	try:
	response = requests.get(url, timeout=30, headers=HEADERS, verify=False)
	response.raise_for_status()
	soup = BeautifulSoup(response.text, "html.parser")

	# Get title
	title_elem = select_element(soup, config["title_selector"])
	title = title_elem.get_text(strip=True) if title_elem else ""

	# Get description
	desc_elem = select_element(soup, config["desc_selector"])
	description = desc_elem.get_text(strip=True) if desc_elem else ""

	# Get content
	content_elem = select_element(soup, config["content_selector"])
	if content_elem:
	paragraphs = content_elem.select(config["content_p_selector"])
	content = "\n\n".join(p.get_text(strip=True) for p in paragraphs if p.get_text(strip=True))
	else:
	content = ""

	# Get publish date - REQUIRED
	date_elem = select_element(soup, config["date_selector"])
	publish_date = date_elem.get_text(strip=True) if date_elem else ""

	# Must have content and publish_date
	if not content or len(content) < 100 or not publish_date:
	return None

	return {
	"source": item["source"],
	"url": url,
	"category": item["category"],
	"title": title,
	"description": description,
	"content": content,
	"publish_date": publish_date
	}

	except Exception:
	return None


	def scrape_source(source: str, config: dict, crawled_urls: set, all_crawled: set):
	"""Scrape articles from a single source."""
	print(f"\n{'='*60}")
	print(f"[{source.upper()}] Starting...")
	print("="*60)

	# Get URLs
	print(f"[{source}] Fetching article URLs...", flush=True)
	article_urls = get_article_urls(source, config, crawled_urls)
	print(f"[{source}] Found {len(article_urls)} new URLs")

	if not article_urls:
	return []

	articles = []
	completed = 0

	def process_article(item):
	return scrape_article(item, config)

	with ThreadPoolExecutor(max_workers=NUM_WORKERS) as executor:
	futures = {executor.submit(process_article, item): item for item in article_urls}

	for future in as_completed(futures):
	item = futures[future]
	completed += 1

	try:
	article = future.result()
	if article:
	articles.append(article)
	with progress_lock:
	all_crawled.add(item["url"])

	# Progress display
	if len(articles) % 10 == 0:
	print(f"[{source}] Progress: {len(articles)}/{TARGET_PER_SOURCE} articles \| Processed: {completed}/{len(article_urls)}", flush=True)

	if len(articles) >= TARGET_PER_SOURCE:
	break
	except Exception:
	pass

	print(f"[{source}] Completed: {len(articles)} articles")
	return articles[:TARGET_PER_SOURCE]


	def main():
	print("="*60)
	print("UVN Dataset Scraper")
	print(f"Target: 6 newspapers x {TARGET_PER_SOURCE} articles = {6*TARGET_PER_SOURCE} total")
	print(f"Workers: {NUM_WORKERS}")
	print("="*60)

	# Load previously crawled URLs
	crawled_urls = load_crawled_urls()
	print(f"Previously crawled URLs: {len(crawled_urls)}")

	all_crawled = set(crawled_urls)
	all_articles = []

	for source, config in NEWSPAPERS.items():
	try:
	articles = scrape_source(source, config, crawled_urls, all_crawled)
	all_articles.extend(articles)

	# Save progress after each source
	save_crawled_urls(all_crawled)

	# Save articles for this source
	source_dir = DATA_DIR / source
	source_dir.mkdir(parents=True, exist_ok=True)
	output_file = source_dir / f"{source}_articles.json"
	with open(output_file, "w", encoding="utf-8") as f:
	json.dump(articles, f, ensure_ascii=False, indent=2)
	print(f"[{source}] Saved to: {output_file}")

	except KeyboardInterrupt:
	print("\nStopped by user. Saving progress...")
	save_crawled_urls(all_crawled)
	break
	except Exception as e:
	print(f"[{source}] Error: {e}")
	continue

	# Save all articles
	all_file = DATA_DIR / "all_articles.json"
	with open(all_file, "w", encoding="utf-8") as f:
	json.dump(all_articles, f, ensure_ascii=False, indent=2)

	print("\n" + "="*60)
	print("SUMMARY")
	print("="*60)
	print(f"Total articles: {len(all_articles)}")
	print(f"Total crawled URLs: {len(all_crawled)}")
	print(f"Saved to: {all_file}")


	if __name__ == "__main__":
	main()